Bybit reported that its security systems helped avert potential losses totaling $700 mln in the first half of 2026. The exchange published an overview of defensive measures and operational metrics to illustrate those results.

Report highlights

During the period, the platform intercepted more than 30,000 suspicious withdrawal requests and thereby protected almost 20,000 user accounts. The company framed these actions as part of a broader security programme combining automation and expert review.

Operational metrics

• Average initial risk assessment: 4.7 minutes, with 95% of checks completed in under ten minutes.
• $212 mln identified as potentially linked to fraudulent activity during investigations performed in the reporting period.
• More than 10,000 malicious addresses were added to the platform blacklist to prevent future interactions.
• Over 100,000 threat alerts were processed with the assistance of artificial intelligence systems.
• Detection of critical vulnerabilities was accelerated by a factor of approximately 3–5 compared with prior periods.

AI and human oversight

The exchange noted that AI tools are used primarily to scale and speed up detection, triage and repetitive workflows across security operations. Final decisions on critical incidents, according to the company, continue to be made by human specialists to ensure contextual judgement and minimise false positives.

Context and next steps

Bybit framed these results as part of ongoing efforts to strengthen user protection through a mix of automated detection and manual review. The exchange did not disclose additional operational details beyond the metrics published in the first-half 2026 security overview.